A campaign procedure of DR (Designated Router), as defined in “RFC4601” standard protocol, is implemented through a PIM Hello message. When a current DR fails, other PIM (Protocol-Independent Multicast) routers may be aware of the failure after timeout of a neighbourship so as to activate a convergence process of multicast routing. However, an interruption time of multicast data transmission due to the process is generally no less than a timeout value of the neighbourship, and thus may be on scale of seconds.
It should be note that, although a unique DR is elected out from any number of PIM interfaces with neighbourship therebetween, only the DRs provided at a source end (that is, so-called first hop node) and an IGMP (Internet Group Management Protocol) joining end (that is, so-called last hop node) would make protocol sense. Herein, the first hop node serves to receive multicast traffic and initiate a registration procedure, and the last hop node serves to process an IGMP joining request, generate (*, G) router and initiate a PIM joining procedure toward upstream.
Chinese patent application No. 200710073938.6 discloses deployment of VRRP (Virtual Router Redundancy Protocol) and BFD (Bidirectional Forwarding Detection) protocols between any related routers, such that convergence efficiency of DR can be improved by triggering a quick election of PIM neighbors through a rapid detection of link failure by means of BFD. Since it is unable to perform the deployment in advance, however, a lot of protocol messages may be introduced into local network, thereby causing a certain waste of network bandwidth. The waste of bandwidth becomes even more obvious especially for a telecommunication scenario of millisecond protection where a high-frequency transmission of detection messages is required.
Chinese patent application No. 201010252067.6 makes improvement based thereon, where a DR and a BDR (backup DR), serving as a backup for the DR, in a shared network segment are elected out by the PIM protocol using a Hello message. The DR takes functions such as distribution or reception of multicast traffic, and a BFD session is only established between the BDR and the DR to monitor survival status of the DR. Once a failure of the DR is detected, the BDR can be switched to a new DR and takes over functions thereof immediately. Since the BFD session is only deployed between DR/BDR, such a solution is more targeted with higher efficiency than that of Chinese patent application No. 200710073938.6.
According to the standard and patent applications described above, the protocol convergence procedure at the time of DR's failure is accelerated by detection of the BFD session between DR/BDR. BDF sessions need to be established between any DR/BDR, including all of interfaces between the DR/BDR, and a millisecond switchover protection can be realized when the sessions are few. As the number of DR/BDR pairs increases, however, the BFD sessions to be maintained increases accordingly. When there are hundreds of or even thousands of sessions, lots of messages may be generated if those sessions down due to a certain reason, such that the time required by switchover may reach second scale and thus protection requirement cannot be satisfied at a telecommunication level. A typical corresponding scenario may refer to an access of Layer-2 network via Layer-3 sub-interfaces of a router (i.e., a deployment of VLAN over the Layer-3 sub-interfaces), in which a common physical link is shared and a same father interface is possessed by those Layer-3 sub-interfaces. When a failure occurs in the physical link, related messages may be generated at all of the Layer-3 sub-interfaces.